def main():
web_dir = os.path.join(opt.results_dir, opt.name,
'%s_%s' % (opt.phase, opt.which_epoch))
webpage = html.HTML(
web_dir, 'Experiment = %s, Phase = %s, Epoch = %s' %
(opt.name, opt.phase, opt.which_epoch))
for i, data in enumerate(dataset):
if i >= opt.how_many:
break
a_parsing_tensor = data['a_parsing_tensor']
b_parsing_tensor = data['b_parsing_tensor']
b_label_tensor = data['b_label_tensor']
b_label_show_tensor = data['b_label_show_tensor']
a_jpg_path = data['a_img_path']
b_jpg_path = data['b_img_path']
pdb.set_trace()
input_tensor = torch.cat(
(a_parsing_tensor, b_parsing_tensor, b_label_tensor), dim=1)
input_var = Variable(input_tensor.type(torch.cuda.FloatTensor))
fake_b_parsing = model.inference(input_var)
test_list = [
('b_label_show', util.tensor2im(b_label_show_tensor[0])),
('a_parsing_tensor_RGB',
util.parsing2im(
util.label_2_onhot(a_parsing_tensor[0],
parsing_label_nc=opt.parsing_label_nc))),
('fake_b_parsing_RGB', util.parsing2im(fake_b_parsing.data[0])),
('b_parsing_tensor_RGB',
util.parsing2im(
util.label_2_onhot(b_parsing_tensor[0],
parsing_label_nc=opt.parsing_label_nc))),
('fake_b_parsing',
util.parsing_2_onechannel(fake_b_parsing.data[0])),
]
print fake_b_parsing.shape
### save image
visuals = OrderedDict(test_list)
visualizer.save_images_parsing_label(webpage, visuals, a_jpg_path[0],
b_jpg_path[0])
print('[%s]process image... %s' % (i, a_jpg_path[0]))
### 从零开始为啥只有12779张？本来12800的！难道有11pair是重复的？检查pair文件。。
### 奇怪哦！难道要12800 + 21
webpage.save()
image_dir = webpage.get_image_dir()
print image_dir
def get_valList(model, opt):
val_list = []
lines = open("./datasets/deepfashion/paper_images/256/val_img_path.txt"
).readlines()
for i in range(len(lines)):
image_a_path, image_b_path = lines[i].split()[0], lines[i].split()[1]
a_jpg_path = os.path.join(opt.dataroot, image_a_path)
b_jpg_path = os.path.join(opt.dataroot, image_b_path)
a_parsing_tensor, b_parsing_tensor, fake_b_parsing, b_label_show_tensor = get_test_result(
a_jpg_path, b_jpg_path, model, opt)
val_list.append(
('val_b_label_show_{}'.format(i), tensor2im(b_label_show_tensor)))
val_list.append(('val_a_parsing_{}'.format(i),
parsing2im(label_2_onhot(a_parsing_tensor))))
val_list.append(('val_b_parsing_{}'.format(i),
parsing2im(label_2_onhot(b_parsing_tensor))))
val_list.append(('val_fake_b_parsing_{}'.format(i),
parsing2im(fake_b_parsing.data[0])))
return val_list
def generate_fake_B(a_image_tensor, b_image_tensor, b_label_tensor,
a_parsing_tensor):
##### stage I #################
input_1_tensor = torch.cat([a_parsing_tensor, b_label_tensor], dim=1)
input_1_var = Variable(input_1_tensor.type(torch.cuda.FloatTensor))
model_1.eval()
fake_b_parsing = model_1.inference_2(input_1_var)
a_parsing_tensor_RGB_numpy = util.parsing2im(
util.label_2_onhot(a_parsing_tensor[0],
parsing_label_nc=opt.parsing_label_nc))
fake_b_parsing_RGB_numpy = util.parsing2im(fake_b_parsing.data[0])
fake_b_parsing_label = util.parsing_2_onechannel(fake_b_parsing.data[0])
a_parsing_RGB_path = './a_parsing_RGB.png'
fake_b_parsing_RGB_path = './fake_b_parsing_RGB.png'
fake_b_parsing_label_path = './fake_b_parsing_label.png'
util.save_image(a_parsing_tensor_RGB_numpy, a_parsing_RGB_path)
util.save_image(fake_b_parsing_RGB_numpy, fake_b_parsing_RGB_path)
cv.imwrite(fake_b_parsing_label_path, fake_b_parsing_label)
##### GEO ######################
theta_json = generate_theta(a_parsing_RGB_path, fake_b_parsing_RGB_path,
geo)
theta_aff_tensor, theta_tps_tensor, theta_aff_tps_tensor = get_thetas_affgrid_tensor_by_json(
affTnf, tpsTnf, theta_json)
theta_aff_tensor = theta_aff_tensor.unsqueeze_(0)
theta_tps_tensor = theta_tps_tensor.unsqueeze_(0)
theta_aff_tps_tensor = theta_aff_tps_tensor.unsqueeze_(0)
#### stage II #################
fake_b_parsing_label_tensor = get_parsing_label_tensor(
fake_b_parsing_label_path, opt)
fake_b_parsing_label_tensor = fake_b_parsing_label_tensor.unsqueeze_(0)
input_2_tensor = torch.cat([a_image_tensor, b_image_tensor, b_label_tensor, a_parsing_tensor, fake_b_parsing_label_tensor, \
theta_aff_tensor, theta_tps_tensor, theta_aff_tps_tensor], dim=1)
input_2_var = Variable(input_2_tensor.type(torch.cuda.FloatTensor))
model_2.eval()
fake_b = model_2.inference(input_2_var)
return fake_b, fake_b_parsing_label_tensor
errors = {
k: v.item() if not isinstance(v, int) else v
for k, v in loss_dict.items()
}
t = (time.time() - iter_start_time) / opt.batchSize
visualizer.print_current_errors(epoch, epoch_iter, errors, t)
visualizer.plot_current_errors(errors, total_steps)
############## Display output images and Val images ######################
if save_fake:
val_list = get_valList(model, opt)
train_list = [
('b_label', tensor2im(b_label_show_tensor[0])),
('a_parsing',
parsing2im(
label_2_onhot(a_parsing_tensor[0],
parsing_label_nc=opt.parsing_label_nc))),
('b_parsing',
parsing2im(
label_2_onhot(b_parsing_tensor[0],
parsing_label_nc=opt.parsing_label_nc))),
('fake_b_parsing', parsing2im(fake_b_parsing.data[0]))
]
val_list[0:0] = train_list
# val_list = train_list
visuals = OrderedDict(val_list)
visualizer.display_current_results(visuals, epoch, total_steps)
### save latest model
if total_steps % opt.save_latest_freq == 0:
print('saving the latest model (epoch %d, total_steps %d)' %
def main():
web_dir = os.path.join(
opt.results_dir, opt.name,
'%s_%s_%s' % (opt.phase, opt.which_epoch, "I_and_II"))
webpage = html.HTML(
web_dir, 'Experiment = %s, Phase = %s, Epoch = %s' %
(opt.name, opt.phase, opt.which_epoch))
for i, data in enumerate(dataset):
if i >= opt.how_many:
break
a_image_tensor = data['a_image_tensor'] # 3
b_image_tensor = data['b_image_tensor'] # 3
b_label_tensor = data['b_label_tensor'] # 18
a_parsing_tensor = data['a_parsing_tensor'] # 1
b_label_show_tensor = data['b_label_show_tensor']
a_jpg_path = data['a_jpg_path']
b_jpg_path = data['b_jpg_path']
##### stage I #################
input_1_tensor = torch.cat([a_parsing_tensor, b_label_tensor], dim=1)
input_1_var = Variable(input_1_tensor.type(torch.cuda.FloatTensor))
model_1.eval()
fake_b_parsing = model_1.inference_2(input_1_var)
a_parsing_tensor_RGB_numpy = util.parsing2im(
util.label_2_onhot(a_parsing_tensor[0],
parsing_label_nc=opt.parsing_label_nc))
fake_b_parsing_RGB_numpy = util.parsing2im(fake_b_parsing.data[0])
fake_b_parsing_label = util.parsing_2_onechannel(
fake_b_parsing.data[0])
a_parsing_RGB_path = './a_parsing_RGB.png'
fake_b_parsing_RGB_path = './fake_b_parsing_RGB.png'
fake_b_parsing_label_path = './fake_b_parsing_label.png'
util.save_image(a_parsing_tensor_RGB_numpy, a_parsing_RGB_path)
util.save_image(fake_b_parsing_RGB_numpy, fake_b_parsing_RGB_path)
cv.imwrite(fake_b_parsing_label_path, fake_b_parsing_label)
##### GEO ######################
theta_json = generate_theta(a_parsing_RGB_path,
fake_b_parsing_RGB_path, geo)
theta_aff_tensor, theta_tps_tensor, theta_aff_tps_tensor = get_thetas_affgrid_tensor_by_json(
affTnf, tpsTnf, theta_json)
theta_aff_tensor = theta_aff_tensor.unsqueeze_(0)
theta_tps_tensor = theta_tps_tensor.unsqueeze_(0)
theta_aff_tps_tensor = theta_aff_tps_tensor.unsqueeze_(0)
#### stage II #################
fake_b_parsing_label_tensor = get_parsing_label_tensor(
fake_b_parsing_label_path, opt)
fake_b_parsing_label_tensor = fake_b_parsing_label_tensor.unsqueeze_(0)
input_2_tensor = torch.cat([a_image_tensor, b_image_tensor, b_label_tensor, a_parsing_tensor, fake_b_parsing_label_tensor, \
theta_aff_tensor, theta_tps_tensor, theta_aff_tps_tensor], dim=1)
input_2_var = Variable(input_2_tensor.type(torch.cuda.FloatTensor))
model_2.eval()
fake_b = model_2.inference(input_2_var)
a_parsing_rgb_tensor = parsingim_2_tensor(
a_parsing_tensor[0],
opt=opt,
parsing_label_nc=opt.parsing_label_nc)
b_parsing_rgb_tensor = parsingim_2_tensor(
fake_b_parsing_label_tensor[0],
opt=opt,
parsing_label_nc=opt.parsing_label_nc)
show_image_tensor_1 = torch.cat(
(a_image_tensor, b_label_show_tensor, b_image_tensor), dim=3)
show_image_tensor_2 = torch.cat(
(a_parsing_rgb_tensor, b_parsing_rgb_tensor,
fake_b.data[0:1, :, :, :].cpu()),
dim=3)
show_image_tensor = torch.cat(
(show_image_tensor_1[0:1, :, :, :], show_image_tensor_2), dim=2)
test_list = [('a-b-fake_b', tensor2im(show_image_tensor[0])),
('fake_image', util.tensor2im(fake_b.data[0])),
('b_image', util.tensor2im(b_image_tensor[0]))]
### save image
visuals = OrderedDict(test_list)
visualizer.save_images(webpage, visuals, a_jpg_path[0], b_jpg_path[0])
if i % 1 == 0:
print('[%s]process image... %s' % (i, a_jpg_path[0]))
webpage.save()
image_dir = webpage.get_image_dir()
print image_dir
model.module.optimizer_D.step()
############## Display results and errors ##########
### print out errors
if total_steps % opt.print_freq == 0:
errors = {k: v.item() if not isinstance(v, int) else v for k, v in loss_dict.items()}
t = (time.time() - iter_start_time) / opt.batchSize
visualizer.print_current_errors(epoch, epoch_iter, errors, t)
visualizer.plot_current_errors(errors, total_steps)
############## Display output images and Val images ######################
if 0:
# valifation for adding skeleton not implemented
val_list = get_valList(model, opt)
train_list = [('b_label', tensor2im(b_label_show_tensor[0])),
('a_parsing', parsing2im(label_2_onhot(a_parsing_tensor[0], parsing_label_nc=opt.parsing_label_nc))),
('b_parsing', parsing2im(label_2_onhot(b_parsing_tensor[0], parsing_label_nc=opt.parsing_label_nc))),
('fake_b_parsing', parsing2im(fake_b_parsing.data[0]))]
val_list[0:0] = train_list
visuals = OrderedDict(val_list)
visualizer.display_current_results(visuals, epoch, total_steps)
### save latest model
if total_steps % opt.save_latest_freq == 0:
print('saving the latest model (epoch %d, total_steps %d)' % (epoch, total_steps))
model.module.save('latest')
np.savetxt(iter_path, (epoch, epoch_iter), delimiter=',', fmt='%d')
# end of epoch